JPH08248964A - Active noise reduction device - Google Patents

Abstract

PURPOSE: To provide a noise reduction device capable of rapidly and surely reducing a noise in space to be muffled without being affected by a noise state and an environmental state such as a temp. and humidity, etc. CONSTITUTION: In the device that a control sound signal Y of a phase opposite to and the same amplitude as a noise signal X detected by a microphone 10 is generated by an ADF 32, and a control sound is generated from a speaker 14, and the filter coefficient W of the ADF 32 is updated successively by an operation part 40 so that a remaining noise signal (e) detected by the microphone 12 becomes minimum, a convergence coefficient operation part 46 successively calculating a convergence coefficient μ for updating the filter coefficient based on the square sum total ΣX<2> of all noise signals is provided. Further, the filter coefficients (transfer functions Gs, Ge) etc., of a DF 34 operating the correction value II of the noise signal X and the DF 38 operating a reference signal R for calculating the filter coefficient are stored in transfer function storage parts 56, 58 at every environmental state beforehand dividing the environmental state of a duct 2, and an environment decision part 54 selects/sets a value according to the real environmental state.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an active noise reduction device for reducing noise in a noise-removed space (for example, a ventilation passage) by sound wave interference between noise and control sound.

[0002]

2. Description of the Related Art Conventionally, as a noise reducing method for reducing noise in an air blowing passage to prevent the noise from leaking into the room from an outlet, a method of attaching a sound absorbing material to an inner wall of the air blowing passage is known. ing. This method is effective for noise in the high frequency range, but does not provide a good noise reduction effect for noise in the low frequency range.

Therefore, in recent years, in order to reduce the noise in the low frequency range, a control sound having the same amplitude as the noise but an opposite phase is actively added to cause interference between the noise and the control sound. Type noise reduction devices have been proposed and are being put to practical use. Further, recently, with the development of a digital signal processor (DSP), it has become possible to use an adaptive digital filter as a filter for generating a control sound having the same amplitude and opposite phase as noise, and in an active noise reduction device. , A system for generating a control sound by adaptive control using such an adaptive digital filter is becoming mainstream (for example, JP-A-5-333877).
No.).

An example of a conventional active noise reduction device that reduces noise by adaptive control using an adaptive digital filter will be described below with reference to FIG.
It should be noted that FIG. 3 is provided in a duct 60 that guides the conditioned air blown from a blower (not shown) to an air outlet (not shown) that is open to the inside of the room. 2) shows the configuration of a conventional active noise reduction device that reduces the noise).

As shown in FIG. 3, in the duct 60, a noise detecting microphone (hereinafter referred to as a noise detecting microphone) 64 for detecting noise in the duct 60 is provided in the duct 60 in order from the upstream side where the conditioned air flows. A speaker 68 for generating a control sound that causes noise to interfere with the background noise, and a residual noise detection microphone (hereinafter referred to as a residual noise detection microphone) 6 for detecting a residual noise after the interference between the control sound and the noise 6
6 is provided.

Further, a sound absorbing material 62 for reducing noise (particularly high frequency noise) is attached to the inner wall of the duct 60, and the microphones 64 and 66 are provided to avoid the influence of the flow of conditioned air. Further, it is covered with the sound absorbing material 62. On the other hand, the sound absorbing material 62 is not arranged at the mounting position of the speaker 68 of the duct 60, and the control sound corresponding to the control sound signal output from the controller 70 is output from the speaker 68.
Radiation can be directly performed inside the duct 60.

Next, the controller 70 generates a control sound signal for generating a control sound from the speaker 68 based on the noise signal and the residual noise signal output from the microphones 64 and 66. An amplifier 7 for amplifying the noise signal and the residual noise signal from the microphones 64 and 66, respectively.
1, 72, analog filters 73 and 74 for removing noise components from the noise signals and residual noise signals amplified by the amplifiers 71 and 72, and the noise signals and residual noises that have passed through the analog filters 73 and 74, respectively. A / D converters 75 and 76 for converting noise signals into digital signals respectively
And a digital signal processing unit 80 for generating a control sound signal (digital signal) for noise reduction based on the noise signal and the residual noise signal converted into digital signals by the respective A / D converters 75 and 76. , This digital signal processing unit 80
D that converts the control sound signal generated in
A / A converter 91, an analog filter 92 for removing noise components from the control sound signal converted into an analog signal by the D / A converter 91, and a speaker 68 for amplifying the control sound signal passing through the analog filter 92. And an amplifier 93 for outputting a control sound from the speaker 68.

Further, the digital signal processing section 80 includes an adaptive digital filter (hereinafter, simply referred to as A) for generating the control sound signal Y.
This AD is mainly composed of 82).
A digital filter (hereinafter, also simply referred to as DF) 86 for generating a correction value H for correcting the noise signal input from the A / D converter 75 based on the control sound signal Y generated in F82; The noise signal is corrected by subtracting the correction value H generated by the DF86 from the noise signal input from the A / D converter 75, and the corrected noise signal X is converted into the ADF8.
Adder 86 for outputting a reference signal to 2 and adder 86
DF88 for generating the reference signal R for updating the filter coefficient of the ADF 82 based on the noise signal corrected in
And a reference signal R and A generated by the DF88.
The filter coefficient calculator 90 sequentially updates the filter coefficient W of the ADF 82 based on the residual noise signal e input from the / D converter 76.

Here, the DF 84 and the adder 86 are arranged so that the noise detection microphone 64 is generated by the control sound generated by the speaker 68.
To prevent howling affecting the
In F84, the control sound signal Y output from the digital signal processing unit 80 is transmitted to the speaker 68 and converted into a control sound, and the control sound is input to the noise detection microphone 64 and returned to the digital signal processing unit 80. Transfer function G of the system up to
By filtering the control sound signal Y based on s (impulse response), the control sound component included in the noise signal input from the A / D converter 75 is obtained as the correction value H, and the adding unit 86 By subtracting this correction value H from the noise signal input from the / D converter 75, the duct 6
The true noise signal X corresponding to the noise within 0 is obtained.

The correction value H is calculated according to the following equation (0). H (n) = Gs ^T · Y (n) (0) where Y (n) = [Y (n), Y (n-1), Y (n-2), ..., Y (n-j) )] ^T ... (0-1) Gs = [Gs0, Gs1, Gs2, ..., Gsj] ^T ... (0-2), and the subscript (n) represents the current value. Further, in the equations (0-1) and (0-2), the left side Y (n) and Gs represent vectors.

Next, the DF 88 and the filter coefficient calculator 9
0 is the filter coefficient W of the ADF 82 so that the residual noise signal e input via the A / D converter 76 is minimized.
To sequentially update. That is, first, regarding the residual noise signal e, the noise signal at the position of the residual noise detecting microphone 66 when the noise propagates through the duct 60 and reaches the residual noise detecting microphone 66 is generated, and the control sound is generated from the speaker 68. When the control sound signal is Y and the transfer function of the system from the digital signal processing unit 80 through the speaker 68 to the residual noise detecting microphone 66 to the digital signal processing unit 80 is Ge, it can be described as the following equation (1).

E (n) = d (n) + Ge ^T · Y (n) (1) where Y (n) = [Y (n), Y (n-1), Y (n-2), , Y (n-j)] ^T (1-1) Ge = [Ge0, Ge1, Ge2, ... Gej] ^T ... (1-2). Also, in equations (1-1) and (1-2), the left side Y (n)
, Ge represents a vector.

On the other hand, the ADF 82 generates an output signal (that is, control sound signal Y) by filtering an input signal (that is, noise signal X) with an FIR filter having an updatable filter coefficient W. , The arithmetic processing is executed according to the following equation (2).

Y (n) = W · X (n) ^T (2) where X (n) = [X (n), X (n-1), X (n-2), ..., X ( ni)] ^T ... (2-1) W = [W0, W1, W2, ..., Wi] ^T ... (2-2). In addition, in the formulas (2-1) and (2-2), the left side X (n)
, W represents a vector.

Here, the filter coefficient W may be sequentially updated according to an adaptive control algorithm so that the control error (that is, the residual noise signal e) is minimized. The adaptive control algorithm may be the steepest descent method. The known Filtered-X LMS algorithm described below can be utilized.

That is, the residual noise signal e (n) is calculated by the above (1)
By substituting the equation (2) into the equation, the following equation (3) can be written. e (n) = d (n ) + Ge T · {W · X (n) T} = d (n) + W T · {Ge · X (n) T} = d (n) + W T · R (n) (3) However, R (n) is a reference signal and can be described by the following equation (4).

R (n) = Ge.X (n) ^T (4) Next, in order to minimize the residual noise signal e (n), an evaluation function J = [e (n)] ^{2 is set} , Considering that the J minimize, J = [e (n) ] 2 = [d (n) + W T · R (n)] 2 ... (5) , and the evaluation function J of the filter coefficient W 2 Since it is a quadratic function, when J is minimum, its derivative ∇
(n) = when ∂J / ∂W becomes zero.

The derivative ∇ (n) can be obtained as in the following equation (6). ∇ (n) = ∂J / ∂W = 2e (n) ・ ∂e (n) / ∂W = 2e (n) ・ R (n) (6) Therefore, the state of the controlled object fluctuates with time. Therefore, in order to optimize the evaluation function J so as to be the minimum corresponding to the state, the filter coefficient W of the ADF 82 may be sequentially updated by using the following equation (7) by the “steepest descent method”. Good.

W (n + 1) = W (n) −μ · ∇ (n) = W (n) −2μ · e (n) · R (n) (7) where μ is the convergence coefficient (step Also referred to as size parameter)
Is. Therefore, the DF88 uses the noise signal X corrected by the adder 86 and the preset transfer function Ge according to the equation (4), and a reference for updating the filter coefficient in the equation (7). The signal R is obtained, and the filter coefficient calculation unit 90 follows the above expression (7), the calculation result (reference signal R), the residual noise signal e input from the A / D converter 76, and the convergence coefficient μ. Based on
The filter coefficient W is sequentially updated.

As described above, according to the conventional active type noise reduction apparatus using the adaptive digital filter, the filter coefficient of the adaptive digital filter is sequentially updated so as to minimize the residual noise in the duct which is the sound dead space. For,
The control sound signal output from the adaptive digital filter converges to a signal capable of sufficiently reducing noise, and theoretically the residual noise can be reduced to zero.

[0021]

However, in such an active noise reduction device, the digital signal processing section 8 is used.
The transfer functions Gs and Ge used in the DF84 and DF88, the filter coefficient calculation unit 90, and the like that form 0, and the convergence coefficient μ
Since the calculation parameters such as are set based on the identification result of identifying the control target by simulation etc., when the device is actually mounted on the control target such as a duct and the control is performed, Due to changes or unexpected input of noise, the calculation parameters may not match the actual control system, and residual noise may not be suppressed satisfactorily.

For example, the convergence coefficient μ used for updating the filter coefficient W is a coefficient for determining the convergence speed of the squared error and the residual error in the successive calculation of the filter coefficient W. If the value is small, the residual noise signal Although e can be converged to the target level, its convergence speed becomes slow,
On the contrary, if the value is large, the convergence speed becomes faster,
If the residual noise signal e becomes large, and if its value is too large, the residual noise signal e cannot be converged to the target level and diverges. Then, as the convergence coefficient μ, if the noise signal X has stationarity and its autocorrelation function is known in advance, the optimum sum of square values of all the noise signals X (sum of squares) is calculated to obtain the optimum value. You can set the value.

However, in order to set the convergence coefficient μ, it is difficult to obtain the average of the sum of squares of all the noise signals X in advance. Therefore, conventionally, a noise state that actually occurs in the controlled object is assumed. Thus, the value of the convergence coefficient μ is determined by thinking and error. Therefore, the convergence coefficient μ does not always become an optimum value depending on the noise state when the control is actually performed, and the convergence speed of the residual noise signal e to the target level becomes slower or the residual noise signal e is sufficiently reduced. There is a problem that it cannot be made smaller.

Further, for example, in the DF84 for calculating the correction value H for preventing howling and the DF88 for calculating the reference signal R for calculating the filter coefficient,
The transfer functions Gs and Ge of the control system are used as filter coefficients, and these transfer functions Gs and Ge are affected by environmental changes such as temperature and humidity. Then, when the usage environment changes from what was assumed at the time of design, the filter coefficients of DF84 and DF88 no longer match the environmental conditions at that time,
Errors occur in the correction value H, the reference signal R, and the like.

That is, when the transfer function Gs of the system from the digital signal processing unit 80 to the speaker 68 to the noise detection microphone 64 to the digital signal processing unit 80 changes due to the environmental change of the control system, the control emitted from the speaker 68. Howling due to sound is likely to occur, and when the transfer function of the system from the digital signal processing unit 80 to the speaker 68 to the residual noise detection microphone 66 to the digital signal processing unit 80 changes, the reference signal R deviates from the optimum value. , The filter coefficient W of the ADF 82 takes a long time to reach the optimum value, the control convergence speed becomes slower, and conversely diverges to deteriorate the noise reduction effect.

The present invention has been made in view of these problems, and is an active noise reduction device capable of promptly and surely reducing noise in a noise-reduced space without being affected by changes in noise conditions and environmental conditions. The purpose is to provide.

[0027]

In order to achieve the above object, the invention according to claim 1 radiates a control sound into a noise-free space to interfere with the noise, thereby reducing the noise in the space. A noise detecting microphone provided in a noise intrusion path from a noise source to a noise-reduced space, and a speaker that receives a control sound signal and emits a control sound into the noise-reduced space. A residual noise detecting microphone for detecting residual noise after interference between the control sound from the speaker and noise, a control sound signal for generating control sound from the speaker, and the noise detecting microphone from the speaker A control sound component in a noise signal input via the noise detection microphone is calculated based on a calculation parameter corresponding to the transfer function of the system, and the noise signal is calculated according to the calculation result. By correcting the noise, the noise signal correction means for calculating a true noise signal corresponding to the noise in the noise-reduced space, and the noise signal obtained by the noise signal correction means are filtered to obtain the control sound signal. An adaptive digital filter, which is an FIR filter capable of updating the filter coefficient, which is generated, the noise signal obtained by the noise signal correcting means, and the residual noise signal input through the residual noise detecting microphone, The filter coefficient of the adaptive digital filter is sequentially updated so as to minimize the residual noise signal based on the set convergence coefficient and the calculation parameter corresponding to the transfer function of the system from the speaker to the residual noise detecting microphone. Filter coefficient updating means for
The sum of squares of all noise signals obtained by the noise signal correction means is obtained moment by moment, and a convergence used for updating the filter coefficients by the filter coefficient updating means using a predetermined arithmetic expression with the sum of squares as a parameter. Convergence coefficient calculation means for sequentially calculating coefficients.

According to a second aspect of the present invention, there is provided an active noise reduction device for radiating a control sound in a noise-reduced space to cause the noise to interfere with the noise and reduce the noise in the space. Similarly to the invention described in (1), a noise detection microphone, a speaker, a residual noise detection microphone, a noise signal correction means, an adaptive digital filter, and a filter coefficient update means are provided, and the noise signal correction means and the filter coefficient update means are further provided. At least one of the above-mentioned calculation parameters used in the above is stored for each calculation of the environmental data such as temperature and humidity in the sound dead space. Means, an environmental state detecting means for detecting an environmental state such as temperature and humidity in the sound dead space, and the above-mentioned performance based on the environmental state detected by the environmental state detecting means. Selecting a calculation parameter that matches the current environmental condition from the calculation data stored in the data storage means, and selecting at least one of the calculation parameters used in the noise signal correction means and the filter coefficient updating means, And a calculation parameter changing unit that changes the selected calculation parameter.

[0029] Next, the invention according to claim 3 is an active noise reduction device for radiating a control sound into a noise-reduced space to cause interference with the noise, thereby reducing the noise in the space.
Similar to the invention according to claim 1 or 2, it comprises a noise detecting microphone, a speaker, a residual noise detecting microphone, a noise signal correcting means, an adaptive digital filter, and a filter coefficient updating means, and further, Residual noise determination means for determining whether or not the residual noise signal input via the residual noise detection microphone has reached a preset target level or less, and the residual noise signal is determined by the residual noise determination means as the target level. If it is determined that the following has been reached, then the update prohibition for prohibiting the update of the filter coefficient by the filter coefficient update means until the residual noise determination means determines that the residual noise signal exceeds the target level. Means and are provided.

According to a fourth aspect of the present invention, in the active noise reduction device according to the third aspect, an environmental condition detecting means for detecting an environmental condition such as temperature and humidity in the noise-removed space, and the environment. Environmental change determining means for determining whether or not the environmental state in the silenced space detected by the state detecting means has changed, wherein the update prohibiting means updates the filter coefficient by the filter coefficient updating means. When prohibited, when the environment change determining means determines that the environmental condition in the sound dead space has changed, the filter coefficient updating means resumes updating the filter coefficient.

On the other hand, the invention according to claim 5 is an active type noise reduction device for radiating a control sound into a noise-reduced space to cause interference with the noise, thereby reducing the noise in the space. Similar to the inventions of Items 1 to 4, a noise detection microphone, a speaker, a residual noise detection microphone, a noise signal correction means, an adaptive digital filter, and a filter coefficient updating means are provided, and the adaptive digital filter further comprises: Control sound signal determination means for determining whether the generated control sound signal exceeds a preset upper limit value, and the control sound signal determination means determines that the control sound signal exceeds the upper limit value, After that, until the control sound signal becomes equal to or lower than the upper limit value, the calculation result by the filter coefficient updating means is multiplied by a correction coefficient smaller than 1, and the filter coefficient of the adaptive digital filter is normally set. A filter coefficient correcting means for correcting a small value Ri, and further comprising a.

Next, an invention according to claim 6 is the active noise reduction device according to any one of claims 1 to 5, wherein a noise detecting microphone, a speaker,
And the residual noise detection microphone is provided in order from the upstream side along the air flow in the path from the blower to the outlet of the air flow path, and each microphone is located on the side surface of the air flow path. It is characterized by being attached to the shade of the formed windshield.

The invention described in claim 7 is the same as claim 6
In the active noise reduction device described in (1), a rectifying component that rectifies the flow of air into a laminar flow state is provided on the upstream side in the vicinity of the mounting position of each microphone in the air flow path.

[0034]

In the active type noise reduction device according to the first aspect of the present invention configured as described above, first, noise that enters the noise-reduced space from the noise source by the noise detection microphone is detected. To be detected. The noise signal from the noise detecting microphone is corrected by the noise signal correction means, and the corrected noise signal is input to the adaptive digital filter.

That is, in addition to the noise from the noise source, the control sound radiated from the speaker is also input to the noise detection microphone. Therefore, if the noise signal from the noise detection microphone is input to the adaptive digital filter as it is. , The adaptive digital filter will generate a control sound signal including not only a signal component for noise reduction having the same amplitude as the noise but an opposite phase, but also a signal component for canceling the control sound generated from the previous speaker. When the control sound is generated from the speaker by the control sound signal, howling of the control sound occurs in the silenced space. Therefore, in order to prevent such howling, in the noise signal correction means, based on the control sound signal generated by the speaker from the control sound, and the calculation parameter corresponding to the transfer function of the system from the speaker to the noise detection microphone, By calculating the control sound component in the noise signal input through the noise detection microphone and removing this control sound component from the noise signal, the true noise signal corresponding to the noise in the noise-reduced space is calculated. Then, this is input to the adaptive digital filter.

Further, the filter coefficient of the adaptive digital filter is preset in the filter coefficient updating means by the noise signal obtained by the noise signal correcting means and the residual noise signal inputted through the residual noise detecting microphone. The residual noise signal is sequentially updated based on the calculated convergence coefficient and the calculation parameter corresponding to the transfer function of the system from the speaker to the residual noise detecting microphone so as to minimize the residual noise signal.
That is, the filter coefficient is sequentially updated according to the Filtered-X LMS algorithm in accordance with the above-mentioned equations (4) and (7).

The convergence coefficient used for updating the filter coefficient is calculated by the convergence coefficient calculation means, and the sum of squares of all the noise signals obtained by the noise signal correction means is obtained every moment.
It is sequentially calculated by using a predetermined arithmetic expression with the sum of squares as a parameter. That is, as described above, if the noise signal has stationarity and its autocorrelation function is known in advance, the convergence coefficient can be set to an optimum value by obtaining the average of the square sums of all noise signals. Therefore, in the present invention, instead of using the convergence coefficient set as a fixed value in advance when updating the filter coefficient, by sequentially calculating the convergence coefficient based on the noise signal, the convergence coefficient that sequentially changes according to the noise signal. Is set and the filter coefficient is updated using this.

As described above, in the present invention, the convergence coefficient used for updating the filter coefficient is sequentially updated on the basis of the noise signal corresponding to the true noise in the silence space obtained by the noise signal correcting means. Therefore, the convergence coefficient can always be set to the optimum value according to the actual noise state, and even if the noise state changes, the control sound emitted from the speaker can be promptly converted into the actual noise. Correspondingly, the residual noise can be promptly and surely converged to the target level or less.

Next, in the active noise reduction device according to claim 2, in place of the convergence coefficient calculation means according to claim 1, calculation data storage means, environmental state detection means and calculation parameter changing means are provided. ing. Then, the calculation parameter changing means changes the calculation data stored in the calculation data storage means to the current environmental condition based on the environmental condition such as the temperature and the humidity in the noisy space detected by the environmental condition detecting means. A suitable calculation parameter is selected, and at least one of the calculation parameters used by the noise signal correction means and the filter coefficient updating means is changed to the selected calculation parameter.

That is, the noise signal correction means uses the operation parameter corresponding to the transfer function of the system from the speaker to the noise detection microphone in correcting the noise signal, and the filter coefficient updating means updates the filter coefficient. In this case, calculation parameters corresponding to the transfer function of the system from the speaker to the residual noise detection microphone are used.However, since the transfer function of the control system changes depending on the environmental conditions such as temperature and humidity in the noise-reduced space, these If the calculation parameter is fixed, the noise signal correction means and the filter coefficient updating means can perform a good calculation operation in the environmental state when the calculation parameter is set, but the environmental state is the environmental state when the calculation parameter is set. , The calculated parameters do not match the actual environmental conditions and There arithmetic errors occur.

Therefore, according to the present invention, when the environmental conditions are divided and the optimum arithmetic parameters are set for each environmental condition and the arithmetic operation is performed by the noise signal correcting means or the filter coefficient updating means. The calculation parameters can be used according to the environmental conditions at that time.

Therefore, according to the present invention, even if the environmental conditions such as the temperature and humidity of the noise-removed space change, the calculation parameters used by the noise signal correction means or the filter coefficient updating means are the actual environment. It is possible to accurately execute noise signal correction or filter coefficient update without being affected by environmental conditions, because there is no case where calculation errors occur in each of these means due to incompatibility with the conditions. become. Therefore, also in the present invention, the residual noise can be promptly and surely converged to the target level or less.

Next, in the active type noise reduction device according to claim 3, the convergence coefficient calculation means according to claim 1 or the calculation data storage means, environmental condition detection means and Instead of the calculation parameter changing means,
The residual noise determining means and the update prohibiting means are provided.

Then, the residual noise determining means determines whether or not the residual noise signal input via the residual noise detecting microphone has reached below a preset target level,
When it is determined by this residual noise determination means that the residual noise signal has reached the target level or less, the update prohibition means
The update of the filter coefficient by the filter coefficient updating means is prohibited until the residual noise determining means determines that the residual noise signal exceeds the target level.

That is, the filter coefficient updating means uses the preset adaptive control algorithm (that is, Filtered-X LMS).
Algorithm), the filter coefficient of the adaptive digital filter is sequentially updated so that the residual noise signal is minimized. When the residual noise signal reaches the target level or less and the residual noise is sufficiently suppressed. Also, if the filter coefficient is continuously updated, the filter coefficient is updated in a direction deviating from the optimum value due to a setting error of various parameters used for updating the filter coefficient, a calculation error at the time of updating, and the like. It is conceivable that the control will be rather deteriorated.

Therefore, in the present invention, when the residual noise signal has reached the target level or less, the filter coefficient is prohibited from being updated by prohibiting the update of the filter coefficient from the current optimum value. I am trying to do it. As a result, according to the present invention, even if the control target is once achieved, the residual noise does not increase and deviate from the control target due to erroneous update of the filter coefficient. You can improve the property.

If the noise condition or the environmental condition changes while the update of the filter coefficient is prohibited, the filter coefficient no longer conforms to the condition and the residual noise exceeds the target level. In this case, since the filter coefficient that has been the optimum value up to now is used as a reference (in other words, the deviation from the optimum value of the filter coefficient is the smallest), the update of the filter coefficient is restarted. The filter coefficient can be quickly converged to the optimum value. Therefore, also in the present invention, the residual noise can be promptly and surely converged to the target level or less.

Next, in the active type noise reduction device according to claim 4, in addition to the device according to claim 3, an environmental condition detecting means for detecting environmental conditions such as temperature and humidity in the noise-removed space. , And environmental change determination means for determining the change in the detected environmental condition. And
When the residual noise signal is below the target level and the update coefficient is prohibited by the filter coefficient updating means from updating the filter coefficient, the environmental change determining means changes the environmental condition in the mute space. If it is determined that the filter coefficient updating means restarts the update of the filter coefficient.

That is, if the environmental condition of the sound dead space changes while the update of the filter coefficient is prohibited, the filter coefficient will deviate from the optimum value, so in the present invention, the environmental condition changes. In this case, even if the residual noise signal is below the target level, the filter coefficient is restarted so that the filter coefficient quickly follows the changing environmental condition.

Therefore, according to the present invention, the filter coefficient can be brought closer to the optimum value more quickly when the environmental change of the sound dead space changes, as compared with the device according to the third aspect. It is possible to further improve the convergence.
On the other hand, in the active noise reduction device according to claim 5, as in the invention according to claims 1 to 4, a noise detection microphone, a speaker, a residual noise detection microphone, a noise signal correction means, and an adaptive digital filter. , And a filter coefficient updating means, as well as a control sound signal determining means and a filter coefficient correcting means.

Then, the control sound signal determining means determines whether or not the control sound signal generated by the adaptive digital filter exceeds a preset upper limit value, and the control sound signal determining means determines the control sound signal. Is judged to have exceeded the upper limit,
After that, until the control sound signal becomes equal to or lower than the upper limit value, the filter coefficient correction means multiplies the calculation result by the filter coefficient update means by a correction coefficient smaller than 1 to correct the filter coefficient of the adaptive digital filter to a value smaller than usual. .

That is, for example, when the noise coefficient changes, the convergence coefficient for updating the filter coefficient does not match the current noise status, and when the value is larger than the optimum value, the filter coefficient updating means The updated filter coefficient, and eventually the control sound signal generated by the adaptive digital filter, also becomes large, the control becomes unstable, and the residual noise cannot be converged below the target level.

Therefore, in the present invention, such a state is judged by whether or not the control sound signal generated by the adaptive digital filter exceeds the upper limit value, and the control signal exceeds the upper limit value and the control becomes unstable. In such a case, the control sound signal generated by the adaptive digital filter is suppressed by correcting the filter coefficient to a value smaller than usual. As a result, according to the present invention, the divergence that occurs when the convergence coefficient is too large is prevented in advance, and the control is stabilized,
The residual noise can be quickly converged to the target level or less.

In addition, as described above, a general active noise reduction device having a noise detection microphone, a speaker, a residual noise detection microphone, a noise signal correction means, an adaptive digital filter, and a filter coefficient updating means is provided. In the apparatus described in claim 1, the convergence coefficient calculation means, in the apparatus described in claim 2, the calculation data storage means, the environmental state detection means and the calculation parameter changing means, and in the apparatus described in claim 3, Is the residual noise determining means and the updating prohibiting means, the apparatus according to claim 4 is the residual noise determining means, the environmental condition detecting means, the environmental change determining means and the updating inhibiting means, and the apparatus according to claim 5 is the The control sound signal determination means and the filter coefficient correction means,
By providing each of them, the stability and the convergence of the control are improved so that the noise in the silenced space can be surely and promptly reduced, and the configuration (claims 1 to 5) If you combine some or all of
It goes without saying that a greater noise reduction effect can be realized by the synergistic action of these means.

Next, an active noise reduction device according to a sixth aspect is a noise detection microphone, a speaker,
A noise reducing device for a blower duct (duct), in which the residual noise detecting microphone is provided in order from the upstream side along the flow of the air in a route from the blower to the outlet of the blower duct through which the air flows. Is. And each microphone is attached to the shade of the windbreak plate formed in the side surface of the ventilation path.

This is similar to the conventional device shown in FIG.
If each microphone is covered with a sound absorbing material, the frequency characteristics of the noise and residual noise detected by each microphone will change from the actual ones, and an accurate noise signal corresponding to the noise and residual noise will be displayed inside the device. Also, it becomes difficult to capture the residual noise signal.

That is, in the field of signal processing technology, an index that represents the correlation between two signals, that is, coherence is defined. The coherence is represented by a value of 0 to 1, and the closer the value is to 1, the higher the correlation and the higher the coherence. In the noise reduction device based on the adaptive control algorithm, the higher the coherence between the noise signal detected by the noise detection microphone and the residual noise signal detected by the residual noise detection microphone, the lower the noise reduction. Is known to be large. Therefore, conventionally, in a device that reduces noise in the air passage in which air flows,
In order to prevent the air in the air duct from directly hitting each microphone, each microphone was covered with a sound absorbing material provided in the air duct. The characteristics may deviate from the actual noise, and the noise reduction effect may not be exhibited satisfactorily.

Therefore, in the present invention, a windshield is attached to the side surface of the air passage, and each microphone is arranged behind the windshield to ensure the coherence of the noise signal from each microphone and to obtain the noise obtained by each microphone. This prevents the frequency characteristics of the signal from deviating from the actual noise. Therefore, according to the present invention, the control sound emitted from the speaker can be controlled according to the actual noise state in the air duct, and in the device according to each of the above claims 1 to 5, in the air duct. It is possible to better reduce the noise.

Next, in the active noise reduction device according to claim 7, in the device according to claim 6,
Further, a rectifying component that rectifies the air flow into a laminar flow state is provided on the upstream side of the position near the mounting position of each microphone in the air passage. This is to prevent the air flowing through the air blowing path from flowing around to the microphones provided behind the windbreak plate and reducing the coherence of the noise signal obtained by each microphone.

That is, for example, when the air passage is formed in a straight line and the air flows straight along the air passage, a large amount of air does not flow into the outside of the windbreak plate provided on the side surface thereof, and Although the microphone is not affected by the air that flows around it, even if the airflow path is curved or the airflow path is straight, the characteristics of the air blower that blows air into the airflow path will cause it to enter the airflow path. When the air flow is disturbed by a large amount, a large amount of air may flow into the outside of the windbreak plate, which may have a great influence on the microphone provided behind it.

Therefore, in the present invention, a rectifying component for rectifying the flow of air into a laminar flow state is provided on the upstream side of the air passage provided with the microphones, thereby preventing the air from flowing around the microphones. -ing As a result, according to the present invention, the coherence of the noise signal obtained by each microphone can be maintained at a large value close to 1, and the noise in the air duct can be reduced more favorably.

[0062]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an active noise reduction apparatus of an embodiment to which all the inventions of claims 1 to 7 are applied. The active noise reduction device of this embodiment is similar to the conventional device shown in FIG. 3 in that the conditioned air blown from the blower 6 of the air conditioning unit is opened to the inside of a room such as an automobile or a train (see FIG. It is provided in the duct 2 leading to (not shown), and it is intended to prevent the noise from leaking into the room from the air outlet by reducing the noise in the duct 2.

As shown in FIG. 1, the active noise reduction apparatus of this embodiment is arranged as close as possible to the blower 6 of the air conditioning unit, which is a noise source, like the conventional apparatus shown in FIG. Microphone for noise detection (noise detection microphone)
10, a speaker 14 for generating a control sound, which is arranged on the downstream side of the noise detection microphone 10 at the shortest distance capable of reducing the noise, and a noise and the control sound which are arranged further downstream of the speaker 14. A residual noise detecting microphone (residual noise detecting microphone) 12 for detecting residual noise after interference and a controller 20 for controlling a control sound emitted from the speaker 14 into the duct 2 are provided.

A sound absorbing material 4 for reducing noise (particularly high-frequency noise) is attached to the inner wall of the duct 2 as in the conventional device. As is clear from the duct sectional view in FIG. Each of the microphones 10 and 12 is not covered with the sound absorbing material 4, and by providing the windbreak plate 9 on the side surface of the duct 2, the outside of the flow path of the conditioned air (that is, the windbreak plate 9
The shade is fixed in the space for the microphone installation.

On the upstream side in the vicinity of the fixed position of the microphones 10 and 12 of the duct 2, a rectifying component having a honeycomb lattice-shaped cross section for rectifying the conditioned air flow in the duct 2 into a laminar flow state. 8a and 8b are provided, respectively. That is, the flow of conditioned air is made laminar by the rectifying components 8a and 8b, and the conditioned air flows around the outside of the windbreak plate 9, and the coherence of the noise signal and the residual noise signal obtained by the microphones 10 and 12 is obtained. Is prevented from falling.

Furthermore, on the wall surface of the duct 2 between the noise detecting microphone 10 and the speaker 14, a temperature sensor 16 and a humidity as environmental condition detecting means for detecting temperature and humidity respectively as environmental conditions inside the duct 2. A sensor 18 is provided, and the controller 20 includes the microphones 10,
In addition to the noise signal and the residual noise signal from 12, the detection signals from these sensors 16 and 18 are also input.

As in the conventional device, the sound absorbing material 4 and the windbreak plate 9 are not provided inside the duct 2 of the speaker 14, so that the control sound can be directly emitted from the speaker 14 into the duct 2. ing. Next, the controller 20
Similar to the conventional device, the amplifiers 21 and 2 for amplifying the noise signals and the residual noise signals output from the microphones 10 and 12 respectively to remove noise components and convert them into digital signals.
2, analog filters 23 and 24, and A / D converter 2
5, 26 are provided. Then, each A / D converter 25,
The noise signal and the residual noise signal converted into digital signals at 26 are input to a digital signal processing unit 30 that generates a control sound signal (digital signal), and the control sound signal generated at this digital signal processing unit 80. Is the D / A converter 4
1, is output to the speaker 14 via the analog filter 42 and the amplifier 43.

The controller 20 has an amplifier 52 and an A / D converter 53 for amplifying the detection signals from the temperature sensor 16 and the humidity sensor 18 and converting them into digital signals in addition to the above-mentioned respective parts similar to those of the conventional device. Equipped with A / D
The detection signal converted into a digital signal by the converter 53 is also
It is input to the digital signal processing unit 30.

Next, the digital signal processing unit 30 is mainly composed of an adaptive digital filter (ADF) 32 for generating the control sound signal Y according to the above equation (2), as in the conventional apparatus. A digital filter (DF) 34 and an addition unit 36 as noise signal correction means for correcting the noise signal by calculating a correction value H for the noise signal input from the D converter 25 according to the expression (0). A reference signal R is generated based on the corrected noise signal X output from the adder 36 according to the equation (4), and the reference signal R and the residual noise signal e input from the A / D converter 26 Based on Equation (7), a digital filter (DF) 38 and a filter coefficient calculation unit 40 as a filter coefficient updating unit that sequentially calculates the filter coefficient W so that the residual noise signal e is minimized. It is equipped with a.

In addition to the above-mentioned respective units similar to those of the conventional apparatus, the digital signal processing unit 30 has a convergence coefficient calculation unit 46, an output level determination unit 48, a noise level determination unit 50, an environment determination unit 54, and a transfer function storage unit. 56 and 58 are provided. Here, the convergence coefficient calculator 46 corresponds to the convergence coefficient calculator of the present invention, and based on the corrected noise signal X output from the adder 36, the total noise signal X at time (n).
The square sum ΣX (n) ² of (n) is calculated every moment, and the square sum ΣX (n) ² and a preset constant α (where 0 <α <
It is 2, and about 0.01 is desirable in the implementation result)
And the tap length L of the filter coefficient of the DF38 (in other words, the transfer function Ge shown in equation (4)) determined at the time of identifying the control system, and the following equation (8) μ = α / {ΣX ( n) ² / L} (8) is used to sequentially calculate the maximum possible value of the convergence coefficient μ, and the filter coefficient calculation unit 40 sets the convergence coefficient μ used for updating the filter coefficient W.

The transfer function storages 56 and 58 correspond to the operation data storage means of the present invention. In the transfer function storage 56, the DF 38 and the convergence coefficient calculation unit 46 include the reference signal R and the convergence. The transfer function Ge used to calculate the coefficient μ and the operation parameter representing the tap length L corresponding to this transfer function Ge are stored for each environmental condition divided into a plurality according to the temperature and humidity of the duct 2. The transfer function storage unit 58 stores the transfer function G used to generate the correction value H for the noise signal in the DF 34.
A calculation parameter representing s is stored for each of the environmental states divided into a plurality according to the temperature and humidity of the duct 2.

The environment determining section 54 corresponds to the arithmetic parameter changing means and the environment change determining means of the present invention. That is, the environment determination unit 54 determines whether or not the environmental condition of the duct 2 has changed, based on the detection signal indicating the temperature and humidity of the duct 2 input from the A / D converter 53,
If the environmental condition has changed, determine whether the current environmental condition corresponds to one of the pre-classified environmental conditions,
The calculation parameter (that is, Ge,
L, Gs) from each transfer function memory 56, 58
38, the convergence coefficient calculation unit 46, and the DF 34, respectively.

On the other hand, the noise level judging section 50 corresponds to the residual noise judging means and the updating prohibiting means of the present invention, and the residual noise signal e inputted from the A / D converter 26 is below a predetermined target level. If the residual noise signal e is less than or equal to the target level, then the residual noise signal e exceeds the target level, or the environment determination unit 54 determines that the environmental condition of the duct 2 has changed. Until that is done, the update of the filter coefficient by the filter coefficient calculation unit 40 is prohibited.

The output level judging section 48 corresponds to the control sound signal judging means and the filter coefficient correcting means of the present invention, and the control sound signal Y generated by the ADF 32.
Is greater than a preset upper limit value, and if the control sound signal Y is greater than the upper limit value, the filter coefficient W of the filter coefficient calculation unit 40 is smaller than 1 by a correction coefficient β ( According to the execution result, β is preferably 0.95 to 0.99), and is corrected to a value smaller than usual.

As described above, in the digital signal processing section 30, various kinds of arithmetic processing for controlling the control sound emitted from the speaker 14 into the duct 2 are executed. Each calculation process is repeatedly executed according to the set procedure. Therefore, next, an execution procedure of each of the above-described arithmetic processes repeatedly executed in the digital signal processing unit 30 as described above will be described with reference to a flowchart shown in FIG.

As shown in FIG. 2, the digital signal processing section 30
When is started, first in S100 (S represents a step), the flags Fa and Fb used in the subsequent processes are reset (0), and an initial setting process for setting initial values to various parameters is executed. To do. Then, in subsequent S110, the detection signals from the temperature sensor 16 and the humidity sensor 18 are read through the A / D converter 53, and in S120, the read detection signal (that is, the current environmental condition of the duct 2) is read. The corresponding transfer functions Gs and Ge and the tap length L are read from the memories as the transfer function storage units 56 and 58, and these respective values are set as calculation parameters used in the subsequent calculation processing.

Thus, in S120, the transfer function Gs,
When a calculation parameter such as Ge is set, the process proceeds to S130, and the latest correction value H calculated in S220 described below is calculated from the noise signal input from the noise detection microphone 10 via the A / D converter 25. Duct 2 by reducing (n)
The arithmetic processing as the addition unit 36 for calculating the true noise signal X (n) in the inside is executed. Since the correction value H (n) has not been calculated immediately after the digital signal processing unit 30 is started, in this case, the correction value H (n) is set to S
The initial value initially set at 100 is used.

Then, in S140 that follows, based on the noise signal X (n) calculated in S130 and the latest transfer function Ge set in S120, the filter coefficient W is updated by using the equation (4). The calculation processing as the DF 38 for calculating the reference signal R (n) of is performed.

Next, in step S150, the sum of squares ΣX (n) ² of the noise signal X (n) at the current time (n), which is a parameter required to calculate the convergence coefficient μ using the equation (8). To S13
Calculated based on the noise signal X (n) calculated at 0, and then S
At 160, the calculated sum of squares ΣX (n) ² and S12
Based on the tap length L set at 0 and the constant α,
The calculation processing as the convergence coefficient calculation unit 46 for calculating the convergence coefficient μ using the equation (8) is executed.

When the convergence coefficient μ is calculated in this manner, it is then determined in S170 whether the flag Fb is reset (0). If the flag Fb is reset, the process proceeds to S180 and the flag is reset. If Fb is set (1), the process proceeds to S210. Incidentally, this flag Fb
Is set when the signal level of the residual noise signal e is determined to be equal to or lower than the target level eo by the processing described later, and the signal level of the residual noise signal e exceeds the target level eo or the environmental condition has changed. The flag Fb is reset at times. That is, in S170, it is determined from the state of the flag Fb whether to update the filter coefficient W.

Next, the flag Fb is reset,
In S180 executed when it is determined that the filter coefficient W is updated in S170, it is determined whether the flag Fa is reset. If the flag Fa is reset, the process proceeds to S190, and vice versa. If the flag Fa is set, the process proceeds to S200. This flag F
The flag a is set in the subsequent processing when it is determined that the control sound signal Y exceeds the preset upper limit value Ymax, and is reset when the control sound signal Y is less than or equal to the upper limit value Ymax. Is. That is, in S180, it is determined from the state of the flag Fa whether the control sound signal Y exceeds the upper limit value Ymax.

Next, the flag Fa has been reset,
In S190, which is executed when the control sound signal Y is less than or equal to the upper limit value Ymax, the current filter coefficient W (n) and S14
The reference signal R (n) calculated at 0 and S160
Of the residual noise detection microphone 12 and the convergence coefficient μ calculated in
The filter coefficient W (n + 1) is calculated (updated) by directly using the equation (7) based on the residual noise signal e (n) input from the A through the A / D converter 26.

On the contrary, in S200 executed when the flag Fa is set and the control sound signal Y exceeds the upper limit value Ymax, the current filter coefficient W (n) and
The reference signal R (n) calculated in S140 and S
Based on the convergence coefficient μ calculated in 160 and the residual noise signal e (n) input from the residual noise detection microphone 12 through the A / D converter 26, the correction coefficient β ( β
The filter coefficient W (n +) is multiplied by the following equation (9) W (n + 1) = β {W (n) -2μ · e (n) · R (n)} (9) multiplied by <1). 1) is calculated (updated). That is, when the control sound signal Y exceeds the upper limit value Ymax, the filter coefficient is calculated by using the equation (9) obtained by multiplying the equation (7) for the filter coefficient W that is normally used by the correction coefficient β. W is corrected to a value smaller than usual.

In this way, when the filter coefficient W is updated in S190 or S200 as the filter coefficient calculation unit 40 or it is determined in S170 that the filter coefficient W is not updated, the process proceeds to S210. , The control output (that is, the control sound signal) Y (using the equation (2) based on the currently set latest filter coefficient W and the corrected noise signal X (n) calculated in S130. calculate n),
The arithmetic processing as the ADF 32 is executed.

Then, in S220, based on the control sound signal Y (n) calculated in S210 and the latest transfer function Gs set in S120, the above equation (0) is used to proceed to S130 next time. The calculation processing as the DF 34 for calculating the correction value H (n) used to calculate the noise signal X (n) is executed.

Next, in S230, a determination process is performed as the output level determination unit 48, which determines whether or not the control sound signal Y calculated in S210 is less than or equal to a preset upper limit value Ymax. If the control sound signal Y exceeds the upper limit value, the flag Fa is set in S240 to update the filter coefficient W in S200, and then the process proceeds to S130. If the sound signal Y is less than or equal to the upper limit value Ymax, the flag Fa is reset in S250 to update the filter coefficient W in S190, and the process proceeds to S260.

At S260, the residual noise signal e input from the residual noise detecting microphone 12 through the A / D converter 26 is inputted.
Of the residual noise signal e detected in step S270, the determination process as the noise level determination unit 50 is performed to determine whether or not the detected signal level of the residual noise signal e is less than or equal to a preset target level eo. Run. If the signal level of the residual noise signal e exceeds the target level eo,
In order to execute the update of the filter coefficient W, the flag Fb is reset in S280, and then the process proceeds to S130, and conversely, the signal level of the residual noise signal e is the target level eo.
If the following is true, the flag Fb is set in S290 to prohibit the update of the filter coefficient W, and the process proceeds to S300.

Next, in S300, the detection signals from the temperature sensor 16 and the humidity sensor 18 are read through the A / D converter 53, and in S310, the read temperature sensor 1 is read.
6 and the detection signal from the humidity sensor 18 (that is, the duct 2
The current environment condition of (1) determines whether or not the calculation parameters such as the transfer functions Gs and Ge and the tap length L have changed to a state where the environment determining unit 54 determines. When it is determined in S310 that the environmental condition has not changed, the process directly proceeds to S130, and when it is determined that the environmental condition has changed, the filter coefficient W is updated normally. S32
At 0, the flag Fb is reset. After resetting the flag Fb in S320, in order to reset the calculation parameter corresponding to the changed environmental state, the above S1
Move to 20.

As described above, in the active noise reduction apparatus of this embodiment, the convergence coefficient μ used in updating the filter coefficient W of the ADF 32 is calculated based on the sum of squares of all noise signals (8 ), The optimum convergence coefficient μ can always be set according to the actual noise state of the duct 2, and residual noise can be promptly and reliably converged below the target level. Can be made.

The tap length L of the filter coefficient of the DF38 and DF34 (that is, the transfer functions Ge and Gs) and the filter coefficient of the DF38 used to calculate the convergence coefficient μ.
Varies according to the environmental condition (temperature / humidity) of the duct 2, but in this embodiment, the filter coefficient and tap length L suitable for each environmental condition are set in advance for each of the environmental conditions divided into a plurality of conditions. DF34, DF3 are stored, and the filter coefficient and tap length L corresponding to the current environmental state are selected from the plurality of filter coefficients and tap length L.
8. The calculation processing by the convergence coefficient calculation unit 46 is executed.
Therefore, according to the present embodiment, the DF 34, the DF 38, and the convergence coefficient calculation unit 46 can obtain an accurate calculation result corresponding to the current environmental state without being affected by the environmental state of the duct 2. Also, the residual noise can be promptly and surely converged to the target level or less.

Furthermore, in this embodiment, when the signal level of the residual noise reaches the preset target level or less, thereafter, the signal level of the residual noise exceeds the target level or the duct 2 is in the environmental condition. The update of the filter coefficient W by the filter coefficient calculation unit 40 is prohibited until the value changes. Therefore, when the residual noise can be reduced to the target level or less, by continuing to update the filter coefficient, the filter coefficient is updated in a direction deviating from the optimum value, which may worsen the control. Rather, the control stability can be improved.

Further, in this embodiment, when the control sound signal generated by the ADF 32 exceeds the preset upper limit value, the filter coefficient W is set to the above value until the control sound signal becomes equal to or lower than the upper limit value. The filter coefficient W is corrected to a smaller value than usual by the correction coefficient β by updating using the equation (9). Therefore, according to the present invention, the divergence of control caused by an erroneous calculation of the convergence coefficient μ is prevented,
Control can improve stability.

As described above, in this embodiment, each parameter used for executing various arithmetic processes in the digital signal processing unit 30 becomes an optimum value corresponding to the noise condition and the environmental condition in the duct 2. Setting, and while monitoring the control results (residual noise, control sound), ADF
Since the updating of the filter coefficient W of 32 is prohibited or the filter coefficient W is corrected, the interior of the duct 2 that receives the conditioned air from the outlet of the duct 2 by the synergistic action of these measures. It is possible to always satisfactorily reduce the noise of the vehicle without being affected by the noise condition and the environmental condition.

Furthermore, in this embodiment, in addition to the improvement of the arithmetic processing in the digital signal processing unit 30, the noise detection microphone 10 and the residual noise detection microphone 12 are provided behind the windshield 9 provided on the side surface of the duct 2. The frequency characteristics of the noise and the residual noise in the duct 2 are changed by fixing and rectifying components 8a, 8b having a honeycomb lattice-shaped cross section for rectifying the flow of air in a laminar state on the upstream side. Instead, the microphones 10 and 12 can be accurately detected. Therefore, according to the present embodiment, the coherence of the noise signal and the residual noise signal input to the digital signal processing unit 30 can be increased, and also by this.
The control accuracy can be improved and the noise reduction effect in the duct 2 can be enhanced.

Although one embodiment of the present invention has been described above, the present invention is not limited to this embodiment.
Various aspects can be taken. For example, in the present embodiment, the device for reducing the noise in the duct 2 that blows the conditioned air from the blower 6 into the vehicle compartment has been described. Even if the device directly reduces the noise in the room by radiating the control sound toward, the present invention can be applied to obtain the same effect as the above.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating a configuration of an active noise reduction device according to an embodiment.

FIG. 2 is a flowchart showing an execution procedure of arithmetic processing in the digital signal processing unit of the embodiment.

FIG. 3 is a block diagram showing a configuration of a conventional active noise reduction device.

[Explanation of symbols]

2 ... duct 4 ... sound absorbing material 6 ... blower 8a, 8
b ... Rectifying component 9 ... Windbreak plate 10 ... Noise detection microphone 12 ... Residual noise detection microphone 14 ... Speaker 16 ... Temperature sensor 18 ... Humidity sensor 20 ... Controller 21, 22, 43, 52 ... Amplifier 23, 24, 42 ... Analog filter 25, 25,
53 ... A / D converter 41 ... D / A converter 30 ... Digital signal processing unit 32 ... ADF (adaptive digital filter) 40 ... Filter coefficient calculation unit 34, 38 ... DF (digital filter) 36 ... Addition unit 46 ... Convergence Coefficient calculation unit 48 ... Output level determination unit
54 ... Environmental determination unit 50 ... Noise level determination unit 56, 58 ... Transfer function storage unit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI technical display location H04R 3/02 G10K 11/16 B

Claims (7)

[Claims] 1. An active noise reduction device for radiating a control sound into a noise-reduced space to interfere with the noise to reduce the noise in the noise-suppressed space, wherein the noise intrusion path from the noise source to the noise-reduced space A microphone for noise detection provided in the speaker, a speaker that emits a control sound in the noise-free space by receiving a control sound signal, and a residual noise detection that detects residual noise after interference between the control sound from the speaker and the noise. Input through the noise detection microphone based on a control sound signal generated by the speaker and a control sound signal generated from the speaker, and a calculation parameter corresponding to a transfer function of a system from the speaker to the noise detection microphone. By calculating the control sound component in the noise signal to be generated and correcting the noise signal according to the calculation result, the true noise signal corresponding to the noise in the noise-eliminated space is calculated. A sound signal correction means, an adaptive digital filter composed of an FIR filter capable of updating the filter coefficient, which filters the noise signal obtained by the noise signal correction means to generate the control sound signal, and the noise signal The noise signal obtained by the correction means, the residual noise signal input via the residual noise detecting microphone, the preset convergence coefficient, and the transmission of the system from the speaker to the residual noise detecting microphone. Filter coefficient updating means for sequentially updating the filter coefficient of the adaptive digital filter so as to minimize the residual noise signal based on the calculation parameter corresponding to the function, and the total noise signal of the noise signal correcting means. The sum of squares is calculated moment by moment, and the filter coefficient updating means uses the predetermined arithmetic expression with the sum of squares as a parameter to filter An active noise reduction device comprising: a convergence coefficient calculation unit that sequentially calculates a convergence coefficient used for updating the number. 2. An active noise reduction device that emits a control sound into a noise-reduced space to interfere with the noise to reduce the noise in the noise-reduced space. A microphone for noise detection provided in the speaker, a speaker that emits a control sound in the noise-free space by receiving a control sound signal, and a residual noise detection that detects residual noise after interference between the control sound from the speaker and the noise. Input through the noise detection microphone based on a control sound signal generated by the speaker and a control sound signal generated from the speaker, and a calculation parameter corresponding to a transfer function of a system from the speaker to the noise detection microphone. By calculating the control sound component in the noise signal to be generated and correcting the noise signal according to the calculation result, the true noise signal corresponding to the noise in the noise-eliminated space is calculated. A sound signal correction means, an adaptive digital filter composed of an FIR filter capable of updating the filter coefficient, which filters the noise signal obtained by the noise signal correction means to generate the control sound signal, and the noise signal The noise signal obtained by the correction means, the residual noise signal input via the residual noise detecting microphone, the preset convergence coefficient, and the transmission of the system from the speaker to the residual noise detecting microphone. Based on a calculation parameter corresponding to the function, the filter coefficient updating means for sequentially updating the filter coefficient of the adaptive digital filter so that the residual noise signal is minimized, and the noise signal correcting means and the filter coefficient updating means are used. At least one of each of the above-mentioned calculation parameters is classified according to environmental conditions such as temperature and humidity in the noise-reduced space. Calculation data storage means for storing calculation data set for each environmental condition, environmental condition detection means for detecting environmental conditions such as temperature and humidity in the noise-removed space, and detection by the environmental condition detection means Based on the environmental conditions
From the calculation data stored in the calculation data storage means, select a calculation parameter suitable for the current environmental condition,
Active noise changing means for changing at least one of the operation parameters used in the noise signal correcting means and the filter coefficient updating means to the selected operation parameter. Reduction device. 3. An active noise reduction device that emits a control sound into a noise-reduced space and interferes with the noise to reduce the noise in the noise-suppressed space. A microphone for noise detection provided in the speaker, a speaker that emits a control sound in the noise-free space by receiving a control sound signal, and a residual noise detection that detects residual noise after interference between the control sound from the speaker and the noise. Input through the noise detection microphone based on a control sound signal generated by the speaker and a control sound signal generated from the speaker, and a calculation parameter corresponding to a transfer function of a system from the speaker to the noise detection microphone. By calculating the control sound component in the noise signal to be generated and correcting the noise signal according to the calculation result, the true noise signal corresponding to the noise in the noise-eliminated space is calculated. A sound signal correction means, an adaptive digital filter composed of an FIR filter capable of updating the filter coefficient, which filters the noise signal obtained by the noise signal correction means to generate the control sound signal, and the noise signal The noise signal obtained by the correction means, the residual noise signal input via the residual noise detecting microphone, the preset convergence coefficient, and the transmission of the system from the speaker to the residual noise detecting microphone. Filter coefficient updating means for sequentially updating the filter coefficient of the adaptive digital filter so as to minimize the residual noise signal based on the calculation parameter corresponding to the function, and the residual noise input via the residual noise detecting microphone. A residual noise determining means for determining whether or not the signal has reached a preset target level or less; If it is determined that the residual noise signal has reached the target level or less, then the filter by the filter coefficient updating means is used until the residual noise signal is determined to exceed the target level by the residual noise determining means. An active noise reduction device comprising: update prohibiting means for prohibiting update of coefficients. 4. The active noise reduction device according to claim 3, wherein an environmental condition detecting means for detecting an environmental condition such as temperature and humidity in the noise-reduced space, and an object detected by the environmental condition detecting means. Environment change determining means for determining whether or not the environmental condition in the muffling space has changed, and the update prohibiting means prevents the update of the filter coefficient by the filter coefficient updating means when the environment is changed. An active noise reduction device characterized in that, when the change determining means determines that the environmental condition in the silenced space has changed, the filter coefficient updating means restarts the update of the filter coefficient. 5. An active noise reduction device that emits a control sound into a noise-reduced space to interfere with the noise to reduce the noise in the noise-reduced space, the noise intrusion path extending from the noise source to the noise-reduced space. A microphone for noise detection provided in the speaker, a speaker that emits a control sound in the noise-free space by receiving a control sound signal, and a residual noise detection that detects residual noise after interference between the control sound from the speaker and the noise. Input through the noise detection microphone based on a control sound signal generated by the speaker and a control sound signal generated from the speaker, and a calculation parameter corresponding to a transfer function of a system from the speaker to the noise detection microphone. By calculating the control sound component in the noise signal to be generated and correcting the noise signal according to the calculation result, the true noise signal corresponding to the noise in the noise-eliminated space is calculated. A sound signal correction means, an adaptive digital filter composed of an FIR filter capable of updating the filter coefficient, which filters the noise signal obtained by the noise signal correction means to generate the control sound signal, and the noise signal The noise signal obtained by the correction means, the residual noise signal input via the residual noise detecting microphone, the preset convergence coefficient, and the transmission of the system from the speaker to the residual noise detecting microphone. Filter coefficient updating means for sequentially updating the filter coefficient of the adaptive digital filter so as to minimize the residual noise signal based on the operation parameter corresponding to the function, and the control sound signal generated by the adaptive digital filter in advance. Control sound signal determining means for determining whether or not the set upper limit value is exceeded, and the control sound signal determining means for controlling sound signal If it is determined that the upper limit value is exceeded, then the calculation result of the filter coefficient updating means is multiplied by a correction coefficient smaller than 1 until the control sound signal becomes equal to or lower than the upper limit value, and the filter coefficient of the adaptive digital filter is calculated. An active noise reduction device comprising: a filter coefficient correction unit that corrects a value smaller than usual. 6. The noise detecting microphone, the speaker, and the residual noise detecting microphone are provided in order from the upstream side along the air flow in a path from the blower of the air flow path to the outlet. The active noise reduction device according to any one of claims 1 to 5, wherein each microphone is attached to a shade of a windbreak plate formed on a side surface of the air passage. 7. The active noise reduction according to claim 6, wherein a rectifying component for rectifying the flow of air into a laminar flow state is provided on the upstream side of the vicinity of the mounting position of each microphone of the air flow passage. apparatus.